This documentation is for an unreleased version of Apache Flink Stateful Functions. We recommend you use the latest stable version.
SDK Appendix #
Stateful Functions can be implemented in any number of programming languages using different language SDKs, both official SDKs provided by the project and 3rd party SDKs maintained by community members. While certain aspects of each SDK API are tailored to feel idomatic to that language a number of basic concepts and traits are universal. This page documents common terms and concepts that exist across all languages.
Function instances are uniquly identified with an
Address which is composed of a
TypeName and identifier. The type name is similiar to a class in an object-oreiented language; it declares what sort of function the address references. The identifier is a primary key, which scopes the funciton call to a specific instance of the function type. When a function is invoked, all actions - including reads and writes of state values - are scoped to the current address.
While some operations simply look at one individual message at a time (for example a parser), many operations remember information across multiple messages. These operations are called stateful.
Some examples of stateful operations:
- When an application searches for certain event patterns, the state will store the sequence of events encountered so far.
- When aggregating events per minute/hour/day, the state holds the pending aggregates.
- When scoring machine learning model, the state holds the current version of the model parameters and information to build feature vectors
- When historic data needs to be managed, the state allows efficient access to events that occurred in the past.
Stateful Functions maintain any number of state values - of various types - and the runtime guaruntees fault-tolerance, reliability, and scalability. State is always scoped to the current address, so multiple function instances of the same type maintain independent values. For example, a greeter type that maintains a seen count will have a unique value for each user.
Stateful Function maintains its own type system to standardize serialization formats across languages and remote execution. Each SDK includes a set of predefined primitive types along with the ability to plug in custom types with your own serializers.
|Boolean||io.statefun.types/bool||A standard boolean type with possible values
|Integer||io.statefun.types/int||A signed 4 byte integer|
|Long||io.statefun.types/long||A signed 8 byte integer|
|Float||io.statefun.types/float||A signed 4 byte floating point number|
|Double||io.statefun.types/double||A signed 8 byte floating point number|
|String||io.statefun.types/string||A UTF-8 encoded string|